Systems and methods that employ a distributional analysis on a query log to improve search results

Abstract

The present invention provides systems and methods that employ a statistical distributional analysis to improve content search engine search results. In particular, a substring and / or a string sequence distributional algorithm can be applied to a set of queries to generate a distributional characteristic (e.g., a profile) for the set of queries, wherein the set is selected from a plurality of queries stored on a query log. Typically, the queries are selected based on a substring of interest and / or an identification of a user initiating searches. The distributional characteristic can then be employed to determine a distributional similarity measure that can be utilized in connection with a search to facilitate search results via providing a mechanism to determine synonymous search terms, spelling corrections / variations, and facilitate collaborative filtering, for example. Thus, the present invention employs a novel technique that mines and employs previous queries to enhance the query search results.

Description

TECHNICAL FIELD[0001]The present invention generally relates to search engine query results, and more particularly to systems and methods that improve content search engine results via a distributional analysis of a search query log.BACKGROUND OF THE INVENTION[0002]The evolution of computers and networking technologies from high-cost, low performance data processing system to low cost, high-performance communication, problem solving and entertainment systems has provided a cost-effective and time saving means to lessen the burden of performing every day tasks such as correspondence, bill paying, shopping, budgeting and information gathering. For example, a computer (e.g., a desktop, a laptop, a hand-held and a cell phone) interfaced to the Internet, via wire or wireless technology, can provide the user with a channel for nearly instantaneous data exchange (e.g., via email, newsgroups and ftp) and merchandise consumption, and access to a wealth of information from a repository of web...

AI Technical Summary

Benefits of technology

[0009]The present invention provides systems and methods that employ a substring and / or a string sequence distributional (e.g., statistical) analysis on a set of search queries obtained from a query log, in connection with a search engine search query, to improve content search engine results. In general, when a user employs a search engine via a web browser to search for information on the Internet, the search string, or query entered into the search engine by the user is typically saved to a log. In addition, information such as a user identification can be saved to the log and associated with the query. Thus, the query log can include executed queries and other valuable information, which can be utilized as a source to learn (e.g., automatically) about a query(s) and a user(s). Such learning can enhance the user's search experience via providing a mechanism to facilitate returning information more pertinent to a respective search query.

[0011]The foregoing can improve search engine query results by providing for synonymous search terms, spelling corrections / variations, and facilitating collaborative filtering, which can enhance subsequent searches in order to return results with a greater degree of correlation to the content being searched by the user. In addition, the statistical pattern can be employed to facilitate determining what queries are failing and why the queries are failing. Furthermore, the statistical pattern can be employed to provide information regarding correlated searches, which can facilitate anticipating a user's needs.

[0016]As noted supra, the foregoing systems and methods can be employed to enhance search engine results via providing a mechanism to determine synonymous search terms, spelling corrections / variations, and to facilitate collaborative filtering. In addition, information indicative of what queries are failing and why queries are failing can be obtained. Moreover, results associated with subsequent searches can achieve a greater degree of correlation to the content being searched.

Problems solved by technology

However, in other instances, the user does not know the URL or the site name.

The large volume of information available via the Internet and the trend to associate a plethora of terms with a site and / or server (e.g., in order to increase the probability of being selected for inclusion in the list of links) commonly results in hundreds or thousands of links returned to the user, wherein many of the links provide access to sites and servers that are not useful to the user.

In other words, it can be observed that the words that precede and follow, or appear in the vicinity of the string “dog” are very similar to the words that appear in the vicinity of the string “cat.” String-based distributional analysis provides a useful mechanism in the running text domain to facilitate determining similar text and improve searches; however, string-based distributional analysis is limited in that it merely employs words that precede and follow the term(s) of interest.

However, such analysis merely analyzes words that precede and follow the strings, and does not exploit the rich information, the search queries and the user information, included in the query log.

Images